Research Policy 2007

+Model
RESPOL 2007 1–19 ARTICLE IN PRESS
Research Policy xxx (2007) xxx–xxx
3 Modes of organizing biomedical innovation in the UK and
F
4 US and the role of integrative and relational capabilities
OO
5 Jacky Swan a,∗ , Anna Goussevskaia a,b , Sue Newell a,c , Maxine Robertson d ,
6 Mike Bresnen e , Ademola Obembe a
7
aWarwick Business School, University of Warwick, Coventry CV4 7AL, UK
8
b Fundação Dom Cabral, Centro Alfa, Av. Princesa Diana 760, Alphaville Lagoa dos Ingleses, Nova Lima, MG, Brazil
9
c Bentley College, 175 Forest Street, Waltham, MA 02452, Boston, USA
PR
10
d Coventry University, Priory Street, Coventry CV1 5FB, UK
11
e University of Leicester Management Centre, University Road, Leicester LE1 7RH, UK
12
13 Abstract
ED
14 Given that biomedical innovation involves intense collaboration across disciplines, occupations and organizations, a nation’s
15 integrative capabilities (the ability to move between basic science and clinical development) and relational capabilities (the ability
16 to collaborate with diverse organizations) have been identified as crucial. This paper deploys qualitative analysis of biomedical
17 innovation in the UK and US to identify mechanisms influencing innovation at the project level through which these macro level
18 capabilities may have effects. From this a propositional framework is developed that helps explain the likely impact of such
19 capabilities for characteristically different kinds of innovation projects at the micro level.
CT
20 © 2007 Published by Elsevier B.V.
21 Keywords: Integrative capabilities; Relational capabilities; Innovation; Organization; Biomedical

22
E
23 1. Introduction ogy to radically change (and hopefully improve) medical 34
treatments and diagnostic techniques, is high. However, 35

RR
24 Biomedical innovation has been defined in various exploitation of scientific breakthroughs for biomedical 36
25 ways but here we see it as a process involving the innovation is problematic, as witnessed by the high cost, 37
26 creation and application of scientific and technological duration and failure rates in product development (CMR 38
27 knowledge to improve the delivery of human healthcare International, 2006). 39
28 and the treatment of disease. This definition is broad Even where scientific knowledge is validated, many 40
CO
29 enough to include new drugs, diagnostics and drug deliv- promising discoveries fail to reach the clinic, with a 41
30 ery regimes for human use, but excludes purely animal, significant number of failures occurring in early devel- 42
31 agricultural and natural resource applications of biotech- opment phases of the innovation process (Hilton et al., 43
32 nology (Rasmussen, 2005). In the biomedical domain, 2002; Dopson, 2005). This is, in part, because biomed- 44
33 the potential for breakthroughs in science and technol- ical innovations – especially more radical innovations – 45
often cut across established professional, occupational 46

UN
and organizational boundaries, and threaten to dis- 47
∗ Corresponding author. Tel.: +44 2476524271;

rupt existing medical practice (Christensen, 2000). This 48
fax: +44 2476524656. suggests that the ability to combine and integrate knowl- 49
E-mail address: jacky.swan@wbs.ac.uk (J. Swan). edge (scientific, technological, commercial, clinical, 50
1 0048-7333/$ – see front matter © 2007 Published by Elsevier B.V.

2 doi:10.1016/j.respol.2007.02.014
Please cite this article in press as: Swan, J. et al., Modes of organizing biomedical innovation in the UK and US and the role of
integrative and relational capabilities, Research Policy (2007), doi:10.1016/j.respol.2007.02.014
+Model
2 J. Swan et al. / Research Policy xxx (2007) xxx–xxx
51 regulatory) across a distributed array of professional the various stakeholders involved (Owen-Smith, 2003; 103
52 groups, commercial organizations, public research orga- Rogers, 1995). For example, how might the ability of 104
53 nizations (PROs) and health organizations would be scientists to move back and forth between basic science 105
54 central to biomedical innovation (Gibbons et al., 1994; and clinical development (i.e. integrative capabilities) 106
55 Coombs et al., 2003). Biomedical innovation processes facilitate an innovation process in a particular project set- 107
56 have thus been described as typically non-linear or ‘inter- ting? In answering this question, we seek to identify the 108
57 active’, comprising complex, uncertain, high risk and specific mechanisms through which macro capabilities 109
58 iterative cycles of knowledge integration and network- are likely to have effects on innovation projects. Within 110
F
59 ing across these diverse groups (Powell et al., 1996; national institutional contexts, there are also clearly wide 111
60 Dodgson et al., 2004). As Powell et al. (1996), note: variations in the ways that innovation projects link- 112
OO
61 “when the knowledge base of an industry is both complex ing public and private science are actually organized 113
62 and expanding, and sources of expertise are widely dis- (e.g. as university start ups, as development projects 114
63 persed, the locus of innovation will be found in networks in biotechnology firms, as R and D projects in global 115
64 of learning, rather than individual firms”. pharmaceutical firms). Our research also seeks to iden- 116
65 The structural features of networks (e.g. their den- tify different types of organizational arrangement for 117
66 sity, scope, strength of ties) linking PROs to commercial biomedical innovation and to explore how these varia- 118
PR
67 firms have been subject to a good deal of analysis (Powell tions in the organization of innovation might mediate the 119
68 et al., 1996; Owen-Smith et al., 2002). This work has impact of a nation’s integrative and relational capabilities 120
69 shown how systematic variation in the composition of at project level. 121
70 networks across nations, and within regions, influences Reflecting the absence of prior work, this study is 122
71 the ability to integrate scientific, clinical and commer- exploratory, inductive and broadly based. This means 123
72 cial expertise (Owen-Smith et al., 2002; Owen-Smith that, whilst we can identify mechanisms at project level 124
ED
73 and Powell, 2004). Of particular importance for biomed- that link to macro capabilities and play an important role 125
74 ical innovation are linkages between scientific research (in our cases) in shaping innovation processes, and have 126
75 and commercial and clinical development. Owen-Smith explored these across contexts (UK, US and different 127
76 et al. (2002) identify two macro-level capabilities that types of project), our data does not allow us to conduct 128
77 influence these linkages: ‘integrative’ and ‘relational’ direct comparative analysis along dimensions defined ‘a 129
CT
78 capabilities. These refer, respectively, to the ability of priori’. That said our findings are suggestive of compar- 130
79 scientists to move back and forth between basic science ative differences and new theoretical propositions which 131
80 and clinical development; and to the ability of organi- future, more deductively oriented work could follow-up. 132
81 zations within an innovation system to collaborate with Our argument is structured as follows. We begin by 133
82 other, diverse organizations. In keeping with ‘national outlining previous literature on the role of integrative and 134
E
83 innovation systems’ approaches, their work suggests that relational capabilities across contexts and attempt to link 135
84 these capabilities stem from macro-level institutional this to innovation processes. We focus in our study on 136
RR
85 differences in the structure and operation of networks the UK and US since, whilst these both have well devel- 137
86 and, thus, differ across nations (Carlsson, 2002; Nelson, oped biomedical industries, previous work has found that 138
87 1993). In particular, Owen-Smith et al. (2002), through they are characteristically different in terms of integra- 139
88 an analysis of ‘upstream’ R and D linkages, have demon- tive and relational capabilities making (to put it crudely) 140
89 strated that these capabilities are better developed in the the US context more supportive of biomedical innova- 141
US institutional context than in Europe (including the tion (Owen-Smith, 2003). However, it should be noted
CO
90 142
91 UK). This, they suggest, accounts (at least in part) for that our study is focused on innovation processes—it is 143
92 the US national advantage in biomedical innovation. not attempting directly to confirm or refute these earlier, 144
93 As yet, however, relatively little research has focused macro findings. 145
94 on identifying and explaining the ways in which such We turn next to the nature of biomedical innova- 146
95 macro capabilities relate to the process of innovation tion processes and suggest that, as these nearly always 147
UN
96 itself. Thus, whilst there is now evidence linking macro entail interactions and interdependencies across spe- 148
97 data on networks to quantitative indicators of innovation cialist knowledge domains and organizations, a more 149
98 (e.g. the development of patents or the diffusion of inno- differentiated framework is needed to understand how 150
99 vations), there is also scope for qualitative research to such processes are organized. Thus, the first phase of our 151
100 complement this by investigating the processes through research study involved an interview-based survey of key 152
101 which macro-level capabilities play out in the expe- stakeholders in the US and UK, combined with review 153
102 riences of actual innovation projects, as perceived by of existing literature, in order to develop a framework 154
+Model
J. Swan et al. / Research Policy xxx (2007) xxx–xxx 3
155 to capture characteristically different modes of organiz- In contrast, relational capabilities facilitate innovation 205
156 ing biomedical innovation projects. This framework is by, for example, supporting collaborative product devel- 206
157 summarized and used as a basis for case selection. We opment projects between PROs, biotechnology firms and 207
158 then focus our analysis on detailed, longitudinal case pharmaceutical firms. The key elements in forming these 208
159 studies in order to identify those mechanisms at project capabilities are linkages across life science networks that 209
160 level that appeared to play an important role in shap- structure national innovation systems. These linkages 210
161 ing innovation processes, and which could be related to have been found to be influenced by the macro institu- 211
162 macro-level capabilities. In keeping with our research tional context, including political (e.g. policy initiatives, 212
F
163 aims, the rationale for selecting characteristically dif- regulation), social (e.g. relations between scientists and 213
164 ferent kinds of project was not direct case comparison technologists, the mobility of the scientific labour force) 214
OO
165 but to capture variation across contexts (Alvesson and and cultural (e.g. values regarding academic participa- 215
166 Skoldberg, 2000). Thus, if a particular mechanism could tion in commercial activities) factors (Owen-Smith et al., 216
167 be identified as relevant across contexts, then we could 2002). 217
168 be more confident of its explanatory value as a basis for The impact of integrative and relational capabili- 218
169 further research. In the concluding section, we attempt to ties on biomedical innovation has been demonstrated 219
170 relate the ways in which integrative and relational capa- through macro-level data comparing the US and Europe 220
PR
171 bilities play out in particular kinds of project through the (including the UK). Thus, Owen-Smith et al. (2002) find 221
172 development of a new propositional framework to guide that the organization of upstream (early stage) R and D 222
173 future research. in the US is qualitatively different to Europe, embrac- 223
ing more diverse constituents and knowledge sources, 224
174 2. Integrative and relational capabilities across closer linkages between basic science and applied sci- 225
175 contexts ence, and closer links between public and private 226
ED
organizations—in the form of dense regionally clustered 227
176 The impact on innovation of national institutional networks (mainly in Boston, California and New York) 228
177 contexts has clearly been the subject of a significant among hospitals, dedicated biotechnology firms, large 229
178 amount of research (Whitley, 2003). We have chosen pharmaceutical firms, universities and research insti- 230
179 in our study to focus on the UK and US because their tutes. Like others, they argue that the US context has 231
CT
180 national systems of innovation have been found in pre- a stronger history of industry-university collaboration in 232
181 vious work to be largely supportive of biotechnology Research and Development (R and D—Rosenberg and 233
182 innovation (Casper and Kettler, 2001; Whitley, 2000; Nelson, 1994) with more, and more diverse, interfaces 234
183 Casper, 2000)—both have world class research facil- between public and private organizations. This reflects 235
184 ities and science bases and internationally recognized national differences in educational systems, career devel- 236
E
185 pharmaceutical firms and both support entrepreneurial opment and labor markets (for example, greater, and 237
186 activity and have active local markets in the supply of more diverse, funding of universities in the US) and the 238
RR
187 technology, scientists and know-how. Recognizing these greater movement of academics back and forth between 239
188 similarities, there are also crucial differences that make PROs, firms and research hospitals (Whitley, 2003). 240
189 them interesting points of contrast. Clark (1987, 2003), With regard to integrative capabilities, these have 241
190 for example, suggests that the UK and US have nationally been linked to national institutional differences in edu- 242
191 distinctive patterns of innovation and distinctive cultural cational systems, career development and labor market 243
repertoires or ‘styles of thinking’ amongst managers, and mobility (Whitley, 2003). For example, in the US con-
CO
192 244
193 the impact of professional and educational institutions text, scientists are more able to move back and forth 245
194 on the legitimation of knowledge has been found to vary between the public and private sector without detriment 246
195 across these contexts (Clark, 2000; Aldrich, 2000). to their scientific careers, as compared to those work- 247
196 In the context of biomedical innovation, Owen-Smith ing in the UK (Owen-Smith et al., 2002; Powell et al., 248
197 et al.’s (2002) delineation of integrative and relational 1996; Mallon et al., 2005). This chimes with work which 249
UN
198 capabilities – capabilities that link scientific research to suggests that the US has a stronger ‘knowledge plus’ ori- 250
199 clinical and commercial development – is particularly entation, where joint goals of understanding and use of 251
200 helpful in understanding the institutional factors that basic science and commercialization are more strongly 252
201 may promote innovation. Integrative capabilities facil- established (Stokes, 1997). This is seen to generate a US 253
202 itate the translation of basic research into commercial advantage in terms of the ability to develop and exploit 254
203 applications through the movement of scientists and their scientific knowledge for clinical development (Owen- 255
204 enhanced labour market mobility (cf. Henderson, 1994). Smith et al., 2002). In contrast, the relationship between 256
+Model
257 PROs and private firms in the UK is more distant: PROs tional capabilities at the macro level (as indicated in 309
258 tend to be more specialized and there is a greater distinc- prior research by the density and scope of network 310
259 tion between basic R and D. These findings concur with ties within regions and nations) would appear to be 311
260 Clark (1987) who observes greater ‘conflict with capital’ related, at project level, to the ability of the organiza- 312
261 in the UK educational system, with a more pragmatic ori- tions involved in innovation processes to acquire and 313
262 entation towards applied, or ‘how to’, knowledge in the create relevant expertise, with informal networks being 314
263 US, as opposed to the stronger value placed on knowl- especially important in this respect. However, further 315
264 edge and understanding for its own sake in the UK. This, research is needed to explore these mechanisms in detail. 316
F
265 he argues generates a polarization in the UK between The value of understanding relationships across dif- 317
266 “academic thinking, which is often regarded as unnec- ferent levels of analysis has also been observed by 318
OO
267 essary and impossible to digest and the rule of thumb Gittell and Weiss (2004) who note that, “frameworks 319
268 empiricism which seems to have a firm grip in many for analyzing organizational phenomena must be respon- 320
269 sectors” (p. 223). Whilst Nowotny et al. (2001) note sive to the dynamic and complex characteristics and 321
270 the increased blurring of boundaries between knowledge inter-relationships between multiple levels of analy- 322
271 traditionally produced in university, government and pri- sis that ‘real life’ situations reflect”. As already seen, 323
272 vate sector research organizations, this is also deeply an important aspect of these dynamics in the case of 324
PR
273 politicized so that “the university appears simultaneously biomedical innovation concerns the ways in which net- 325
274 as capturing, but also captured” (p. 79). worked relations and the combination of specialist forms 326
275 With regard to relational capabilities, there is ample of knowledge and expertise are actually coordinated 327
276 evidence that individual firms, even large pharmaceu- amongst the different parties involved in innovation 328
277 ticals, do not possess all of the resources necessary to projects. In the analysis below, then, we begin by iden- 329
278 successfully develop new therapeutics (Powell et al., tifying characteristically different modes of organizing 330
ED
279 1996). Individuals and firms, therefore, need to collab- biomedical innovation projects (focusing on network 331
280 orate formally and informally to acquire the necessary relations and knowledge flows), before looking across 332
281 resources. In the biotechnology sector, ‘open’ channels different kinds of project in the UK and US to draw out 333
282 have been found to be particularly helpful in facilitat- those mechanisms that appear to be important in relating 334
283 ing opportunities for knowledge creation through the macro capabilities to actual innovation processes. 335
CT
284 enhanced likelihood of ‘spillover effects’: that is, knowl-

285 edge is more likely to ‘leak’ though more open channels 3. Methodology 336
286 (Owen-Smith and Powell, 2004; Murray, 2002; Kreiner

287 and Schultz, 1993). Similarly, Salman and Saives (2005) The findings below are drawn from a 3-year 337
288 show that, as well as being influenced by direct rela- exploratory study of innovation in the UK and US 338
E
289 tionships between organizations, innovation outcomes biomedical sectors aimed at identifying the factors 339
290 are attributable to informal, unpredictable relationships facilitating and impeding innovation projects across 340
RR
291 generated through indirect ties, which create access to contexts. The UK and US contexts were selected in 341
292 expertise beyond formal alliance partnerships. Other lit- order to achieve broadly defined variation (Alvesson 342
293 erature has emphasized the importance of trust-based, and Skoldberg, 2000) as earlier work had suggested that 343
294 informal networks for R and D (e.g. Liebeskind et al., macro level relational and integrative capabilities were 344
295 1996; Kreiner and Schultz, 1993). The US biomedical different across these two contexts (Owen-Smith et al., 345
sector has also been found to have better established rela- 2002).
CO
296 346
297 tional capabilities than in Europe reflected, for example, The first phase was an interview-based survey with a 347
298 in the scope and density of networks—at least in certain range of individuals representing key stakeholder groups 348
299 regions. who had significant experience of working in early-stage 349
300 This prior research is indicative of the ways in which biomedical innovation projects that could be described 350
301 integrative and relational capabilities might influence as involving ‘systemic production networks’ (Alter and 351
UN
302 innovation processes in broad terms. For example, it Hage, 1993). Interviewees were initially identified via 352
303 suggests that one important mechanism via which inte- members of our project’s expert UK and US Scientific 353
304 grative capabilities could influence innovation at project Advisory Boards (SAB). Members of the SAB were 354
305 level is through the career identities and values of all experienced in managing biomedical innovation, 355
306 individual scientists involved in project work, in par- and included serial entrepreneurs of biotech compa- 356
307 ticular the extent to which they see goals of science nies, venture capitalists, academic scientists and policy 357
308 and commerce as mutually acceptable. Similarly, rela- specialists. From these initial contacts, additional inter- 358
+Model
359 viewees were identified using a ‘snowballing’ technique. socially constructed by those involved. For example, 411
360 This kind of non-probability convenience sampling is in some projects, the acquisition of funding constituted 412
361 appropriate when the research is exploratory and popu- ‘success’ when this facilitated future development work 413
362 lation parameters are unknown (Saunders et al., 2000). and enabled progress to be made on projects. In oth- 414
363 We conducted 97 interviews (44 UK; 53 US). In ers, clinical trial results which secured gateway FDA 415
364 addition, 22 meetings were held to discuss further par- approval served as an important proxy for success by 416
365 ticipation as case studies for phase two (17 UK; 5 those involved. The length of time available to study 417
366 US). Recognizing that there are regional variations, these projects (@30 months) was adequate to trace what 418
F
367 field work focused on the Boston area in the US, progress (if any) had been made on projects and what fac- 419
368 which has a concentration of biomedical-related orga- tors may have facilitated or hindered projects. Thus, on 420
OO
369 nizations (including Harvard, MIT and Mass General, each case visit (held approximately 6 monthly) project 421
370 for example). In the UK, we concentrated on the team members were asked to review whether progress 422
371 Oxford–Cambridge–London triangle, which is also rec- had matched their expectations in the prior period and 423
372 ognized for its high level of activity and reputation for also to indicate their expectations for the forthcoming 424
373 innovation in the biomedical area. The primary aim period. However, the time available was not sufficient to 425
374 of these first phase interviews was to gather rich, and assess ‘success’ in definitive terms as the entire devel- 426
PR
375 broadly representative, descriptions of experiences of opment process would take around 8 years on average 427
376 innovation in the biomedical field from those directly (CMR International, 2004). This collective case study 428
377 involved. In particular, we wanted to establish the differ- approach was aimed at facilitating interpretation and 429
378 ent ways biomedical innovation projects were organized. developing qualitative insights into the early-stage devel- 430
379 Whilst the dimensions we identified (organizational cou- opment process (Alvesson and Skoldberg, 2000), by 431
380 pling and knowledge boundaries, see below) do resonate comparing the similarities and differences provided by 432
ED
381 with earlier work (Alter and Hage, 1993), this ear- multiple settings – in this case – projects (Tsoukas, 433
382 lier work had not specifically considered biomedical 1989). 434
383 projects, or the relationship between knowledge flows Whilst access was negotiated via focal organizations 435
384 and networks in systemic production networks which and/or individuals, the unit of analysis was the innova- 436
385 relational and integrative capabilities directly relate to. tion process, as manifest in particular projects over the 437
CT
386 The second phase comprised longitudinal case stud- period of the research, not a specific firm, so intervie- 438
387 ies of innovation projects (six in the US and four in the wees spanned the different organizations involved. The 439
388 UK), offering exemplars of different ways of organiz- cases were selected on the basis of, first, the choice of 440
389 ing biomedical innovation. The particular focus was on research topics and questions being posed (Stake, 1995) 441
390 early development (the move from discovery to the early and, second, the possibility of capturing both historic 442
E
391 stages of commercialization) as the first phase identi- and ‘live’ processes to inform the longitudinal analysis 443
392 fied this as a critical point at which upstream scientific (Pettigrew, 1990). Thus, in all cases, activity relating to 444
RR
393 expertise needed to interact with other forms of expertise the innovation process had been on-going for at least 2 445
394 (clinical, commercial, regulatory) and where integrative years and was projected to continue during the research 446
395 and relational capabilities would be expected to have period. This permitted the collection of data providing 447
396 a significant impact (Owen-Smith et al., 2002). A key current, as well as retrospective, views of the innovation 448
397 criteria for case selection was, not only that the case process. This case research was interview based, with 449
appeared to offer a good example of a particular mode a minimum of four fieldwork visits per case over the
CO
398 450
399 of organizing as identified by phase 1, but also that 30-month period. On average, 14 interviews were con- 451
400 good access to key stakeholders was offered, allowing us ducted per case. Interviews were complemented with 452
401 to conduct detailed longitudinal research including, for access to extensive documentary data (including com- 453
402 example, allowing investigators non-participatory obser- panies’ reports, inter-partner correspondence, contracts 454
403 vation of strategy and project meetings and access to and meeting minutes) and observational data (including 455
UN
404 meeting notes. non-participant observation of project team and advisory 456
405 Our approach was rooted in social constructivism board meetings). The interviews in both phases were 457
406 (Kuckla, 2000) and therefore a subjective rather than recorded and transcribed. Detailed notes were taken and 458
407 an objective epistemology was assumed (Denzin and later written up as a record of any meetings attended. 459
408 Lincoln, 1998). Hence, we did not aim to compare cases The first phase data were coded and analyzed using 460
409 along fixed dimensions such as ‘success’ of projects, NVivo software and the ‘memoing’ technique (Glaser, 461
410 as we accepted that what counted as ‘success’ was 1978; Miles and Huberman, 1994). Interviewees were 462
+Model
463 asked to recount stories of their own experiences of of both US and UK contexts). We explicitly sought from 515
464 biomedical innovation projects and to talk about the our advisory board members, feedback as to how far the 516
465 factors and critical events that had influenced these categories, models and frameworks that we were devel- 517
466 projects. In the analysis, we focused on identifying the oping, resonated with their own experiences in the field. 518
467 institutional level factors that interviewees discussed as Their validation of our analyses provided further support 519
468 either facilitating or impeding the innovation process for our findings. 520
469 described. This included, for example, access to finance,

470 availability of expertise and access to technology. We 4. Modes of organizing biomedical innovation 521
F
471 used this analysis to consider how institutional-level rela-
472 tional and integrative capabilities were playing out in From the first-phase analysis, coupled with literature 522
OO
473 different innovation projects. review, two broad dimensions were found to be help- 523
474 In phase two, recognizing the complexity of the cases, ful in characterizing innovation projects. We refer to 524
475 the research for each case was jointly conducted by at these here as ‘organizational coupling’ and ‘knowledge 525
476 least two investigators in order to co-develop interpreta- boundaries’. These dimensions are depicted in Fig. 1 526
477 tions in real time throughout the research period. Data and described briefly next. Details on the derivation of 527
478 were managed, coded and recoded, using NVivo, the ini- these dimensions from the phase 1 data can be found
PR
528
479 tial coding structure having been developed from phase elsewhere (Swan et al., 2005). 529
480 1. On completion of fieldwork, detailed case descriptions Organizational coupling refers, then, to the organi- 530
481 were produced by the investigators on each case (on aver- zation and management of collaborations and network 531
482 age 10,000 words) containing primary data (quotes from ties between partners. Variation along this dimension 532
483 interviews, inserts from documents, etc.), and structured ranged from networked and loosely coupled modes, 533
484 according to particular themes. All case descriptions to more hierarchical and tightly coupled modes (cf. 534
ED
485 were then content analyzed by the whole team in order Alter and Hage, 1993; Hardy et al., 2003; Owen-Smith 535
486 to identify the processes and project dynamics that facil- and Powell, 2004). In the former, innovation projects 536
487 itated or impeded project progress (however, defined by were pursued by partners joined in a loosely coupled 537
488 project participants), relating each to either the ability network of organizations, with work being conducted 538
489 of scientists to move back and forth between basic sci- across several organizations, each of which had signif- 539
CT
490 ence and clinical development (i.e. linked to macro-level icant autonomy in the management of their own work. 540
491 integrative capabilities), or to the ability of organiza- Allocation of resources to tasks (including people, equip- 541
492 tions within an innovation system to collaborate with ment and finances), where discussed, was negotiated 542
493 other, diverse organizations (i.e. linked to macro-level through largely informal means. Thus, management in 543
494 relational capabilities). We refer to these processes and such projects was decentralized and vertical dependency
E
544
495 dynamics for convenience as ‘mechanisms’ and the on centralized resources was low (Alter and Hage, 1993). 545
496 analysis below explains why and how we see these mech- Commitment to tasks was based primarily on mutual 546
RR
497 anisms as related to macro capabilities.

498 Initially, 12 mechanisms were identified, which
499 related to integrative capabilities and 17 mechanisms,
500 which related to relational capabilities. Further inductive
501 analysis of these 29 mechanisms by each team mem-
ber and then by the whole team led to a clustering of
CO
502
503 these into 8 primary mechanisms identified from the 10

504 cases. The major criteria by which interpretive research
505 is assessed – trustworthiness, credibility, confirmability
506 and transferability – were therefore addressed by inde-
507 pendent verification across the 10 case studies and the 6
UN
508 researchers on the project (Denzin and Lincoln, 1998).

509 In addition, over the duration of the research, five SAB
510 meetings were also held. At each of these meetings,
511 major findings to-date were discussed, including, at the
512 final meeting, a detailed discussion to test our analysis
513 of how the mechanisms we identified linked to macro
514 capabilities (most SAB members having had experience Fig. 1. Modes of organizing biomedical innovation.
+Model
547 trust and obligation to a super-ordinate goal (e.g. pro- transformation and blending of knowledge, expertise 599
548 ducing a breakthrough new treatment) and promises of and artifacts across existing disciplines and occupations 600
549 expected return, rather than on detailed or formally bind- (Bock et al., 2003). High knowledge boundaries also 601
550 ing contracts. Where formal contracts did exist, these arose in situations where the parties involved needed to 602
551 centred broadly on mutual obligations and the allocation combine their expertise, in order to articulate new knowl- 603
552 of future financial gains (e.g. split of revenues gener- edge and practices, but had not worked together before 604
553 ated). Knowledge flows could be described as occurring or, indeed, with others in those particular disciplines or 605
554 via relatively ‘open channels’, characterized by diffuse occupations—for example, when each organization had 606
F
555 linkages and ‘knowledge spillovers’ (Owen-Smith and a ‘piece’ of technology, or IP, and only the combina- 607
556 Powell, 2004). tion of the pieces would allow the development of the 608
OO
557 In contrast, in tightly coupled modes, most innova- product. These kinds of situation engendered ‘pragmatic 609
558 tion activity was carried out and coordinated within a boundaries’ (Carlile, 2004), meaning that alignment of 610
559 large focal firm but with clearly identified parts of the professional interests and development of shared expec- 611
560 work (e.g. manufacturing, clinical trials) being formally tations among stakeholders was crucial. 612
561 contracted to other parties. Management was relatively The combination of these two dimensions provided 613
562 hierarchical and there was high dependency on central- a novel framework for understanding characteristi- 614
PR
563 ized resources, knowledge flows occurring via closed cally different modes of organizing innovation projects, 615
564 ‘conduits’, or pipelines (Owen-Smith and Powell, 2004). depicted in the four quadrants of Fig. 1. In the next 616
565 Legally binding contracts existed to secure deliverables section, these different kinds of innovation project are 617
566 and protection of IP and detailed financial responsi- described and illustrated through case descriptions. 618
567 bilities and returns as well as the allocation of tasks, Whilst our analysis, as seen, drew from all 10 cases, 619
568 deadlines, roles responsibilities, risk management and we have elected, due to pressures on space, to focus 620
ED
569 so forth. here on 6 to illustrate these (3 UK, 3 US). We also 621
570 The knowledge boundaries dimension relates specif- focus on the projects represented by quadrants I, III and 622
571 ically to ways in which knowledge was combined across IV. Whilst we collected ‘vignettes’ of quadrant II-type 623
572 the different specialist domains involved and ranges projects from the first phase, our cases studies were not 624
573 from ‘high’ to ‘low’. It is important to emphasize here concentrated here because these examples were typi- 625
CT
574 that, in keeping with a constructivist approach, we saw cally centred on more traditional, incremental innovation 626
575 knowledge domains not as purely cognitive, but as tied processes contained within the R and D departments of 627
576 to boundaries of specialized practice (Carlile, 2004; large global pharmaceutical firms, with technology being 628
577 Orlikowski, 2002). Nearly, all biomedical projects dis- either developed in-house, bought or licensed. Whilst 629
578 cussed in phase 1 deployed multidisciplinary teams these kinds of innovation project are interesting in their 630
E
579 and so spanned cognitive domains—different forms of own right, they are considered less relevant to our main 631
580 knowledge (scientific, commercial, clinical, regulatory) area of interest (i.e. the link between innovation pro- 632
RR
581 clearly needed to be brought together in develop- cesses at project level and a nation’s capabilities). 633
582 ment (Howard-Grenville and Carlile, in press; Gay and

583 Dousset, 2005). The important issue here, then, was not 4.1. Quadrant I cases 634
584 simply multidisciplinarity, but whether or not the work

585 involved in innovation projects actually demanded new Quadrant I, characterized by loosely coupled orga- 635
ways of practicing across these domains that meant that nizational relationships and low knowledge boundaries,
CO
586 636
587 existing knowledge/practice boundaries had to be over- was typically populated by small early stage spin-off 637
588 come. companies started by academics with entrepreneurial 638
589 High knowledge boundaries arose in situations where interests and/or commercial experience. There was high 639
590 there were high novelty areas involved, where medical dependency on the parent university or PRO (Powell et 640
591 need was ambiguous and/or contested and where impli- al., 1996) and multiple sources of funding were sought 641
UN
592 cations for clinical practice were difficult to forecast for facilities, equipment, consumables and specialist 642
593 (Carlile, 2004). In such projects, the intensity of knowl- expertise (e.g. from research grants, venture capital 643
594 edge sharing between those involved in upstream science investors and larger biotechnology or pharmaceutical 644
595 (e.g. scientific research) and those involved in down- firms). However, the development of the science and 645
596 stream application (e.g. clinical practice) was greater. An technology required relatively low levels of knowledge 646
597 example of such a situation was the development of tis- integration across those groups involved, knowledge 647
598 sue engineered products, which required the progressive being produced in sub tasks performed more or less 648
+Model
649 separately and pooled subsequently (Thompson, 1967). ech had a loosely coupled management structure, relying 701
650 Innovations here did promise significant improvements on a number of part-time executives and administrative 702
651 in treatment, but were minimally disruptive to existing personnel, coordinated by one of the two founding sci- 703
652 modes of treatment delivery and so did not depend cen- entists (who continued their academic activities at the 704
653 trally on combining the production of the science or university). Although, the academic Director had a his- 705
654 technology with the particular expertise of clinicians that tory of interaction with industry, via involvement on the 706
655 would likely deploy it (Christensen, 2000). advisory boards of biotechnology companies, the com- 707
656 NewPharma was a start up based in the US developing pany still relied heavily on the technology transfer office 708
F
657 a new therapeutic for a neurological disease based on the of the university, which provided access to biotech com- 709
658 founding academic’s discovery in his Hospital lab. The panies and the VC community. 710
OO
659 company was established with personal funding from SampaTech’s first CEO had formerly worked as a 711
660 the founders, the scientist who made the original discov- manager in a large pharmaceutical firm but was later 712
661 ery and an ‘entrepreneurial’ partner who took the role considered to lack the scientific background required to 713
662 of CEO. The three had previous experience of launch- secure further VC funding. A new CEO was appointed, 714
663 ing a company together. The founding scientist did not with the help of the technology transfer office, who had 715
664 take an operational role in the company but served as ‘the right’ profile. She had previously started several 716
PR
665 a member of the management and scientific advisory biotechnology companies thereby developing a reputa- 717
666 boards. Whilst he viewed himself as a ‘risk taker’, there tion as a ‘serial entrepreneurial scientist’. This change of 718
667 was an express awareness of the limits to which he could CEO resulted in a major refocusing of the organization 719
668 engage in commercialization activities and still retain his strategy, which was initially geared towards the devel- 720
669 academic position. opment of three technology platforms, thereby requiring 721
670 NewPharma obtained a licensing option from the extensive funding. By reassessing the strategy, the new 722
ED
671 Hospital (who held the patent) with a view to following CEO narrowed the area of development to therapeutics, 723
672 one of two options: license their compound to a phar- thereby generating external pharmaceutical interest in 724
673 maceutical company or develop it within NewPharma. licensing the lead product. 725
674 They delivered presentations to senior pharmaceutical

675 executives and research scientists with links to pharma- 4.2. Quadrant III cases 726
CT
676 ceuticals. Where individuals appeared to be interested,

677 they were invited to help NewPharma negotiate licens- The projects in quadrant III, like those in quadrant II, 727
678 ing deals with pharmaceutical companies with which were usually led by larger biotechnology or pharmaceu- 728
679 they had connections, in return for shares in the com- tical firms. However, whereas quadrant II, projects were 729
680 pany. These ‘deal breakers’ were essential to get access aimed at incremental improvement of currently available 730
E
681 and credibility with potential investors in pharmaceutical treatments, quadrant III describes cases where the com- 731
682 companies. Simultaneously, there was a need for valida- panies ventured into highly innovative areas where the 732
RR
683 tion of the discovery by academic peers. This was to be development of breakthrough technologies placed high 733
684 achieved by submitting a paper to a reputable journal. demands on the focal organization to collaborate with 734
685 During the period of the study, NewPharma was unsuc- basic researchers and, given their potential to disrupt 735
686 cessful in getting investment. Based on feedback they medical practice, also required constant interaction with 736
687 received from the presentations that expressed concern (and input from) end users (health professionals) and 737
that a single discovery was not a sufficient platform to regulatory bodies. These innovation projects were man-
CO
688 738
689 attract investment, NewPharma continued to search for aged centrally, based on formal contractual agreements 739
690 a partner with additional patents to build a platform. with smaller companies and specialist research organiza- 740
691 SampaTech was a small company developing novel tions in ways reminiscent of supply chain interactions. 741
692 therapeutics for hepatitis. It was founded by two sci- However, whilst inter-organizational relationships and 742
693 entists from a leading UK university, who developed financial resources were tightly controlled by the cen- 743
UN
694 the basic technology in collaboration with a large phar- tral organization, knowledge was widely distributed and 744
695 maceutical firm (that subsequently withdrew from the needed to be brought together across different domains 745
696 project) and another university. By early 2005, the com- of practice. 746
697 pany had acquired two rounds of seed funding: one from AmericanBio was a relatively large US biotechnology 747
698 one of the universities and another from a donation. The company. ELBOW was a product for cartilage repair 748
699 intention was to develop the lead project, out-license it based on tissue engineering technology. The ELBOW 749
700 and use the royalties for further developments. SampaT- project was conducted by a multifunctional core team. 750
+Model
751 Importantly, AmericanBio had a strong internal regula- and there was always a risk that the results might not 803
752 tory group responsible for interaction with the FDA. This prove favourable. 804
753 regulatory expertise was crucial for ELBOW, since ini- Pharmaceutical companies who expressed an interest 805
754 tially there was no regulatory framework and American- in partnering were given an information pack about the 806
755 Bio were able to shape the regulations to secure approval project produced by Body’s project team. Whilst these 807
756 for their first generation product. Besides the challenges firms were conducting their own due diligence projects, 808
757 involved in regulation, sales and marketing had also they were also expected to deliver ‘capability presenta- 809
758 proven costly and complex as the product disrupted tions’ outlining the sorts of trials they planned to conduct 810
F
759 established ways in which orthopedic surgeons (the main to progress the project and the resources they would 811
760 users) practiced. Thus, development required a signifi- commit. This was because Body wanted to be confi- 812
OO
761 cant degree of interaction with the user community. dent that the partner would have sufficient knowledge 813
762 Currently, AmericanBio is developing a new genera- and resource to move forward with ANTIBODY-2 and 814
763 tion of ELBOW. Setbacks with its internal development take it to market quickly. Within Body, there was a lot of 815
764 prompted the decision to search for external technol- informal personal networking across the various teams 816
765 ogy that could help ‘leapfrog’ the project through early involved and with outsiders—as one respondent noted 817
766 stage clinical trials. The company had a special inter- ‘everybody in this industry knows everybody’. However, 818
PR
767 est in EU companies, because the lack of regulation of the decision was made to exclude from partnering dis- 819
768 tissue engineered products in Europe meant that patient cussions anyone involved in existing relationships with 820
769 data on the technology was available that might help the candidate companies. In this way, formal networks 821
770 ease the progression to clinical trials in the US (inter- were emphasized and information management during 822
771 estingly, while AmericanBio had played a major role in due diligence was deemed crucial. 823
772 shaping US regulation, which made it difficult for com-

ED
773 petitors to enter the market, they now had to face those 4.3. Quadrant IV cases 824
774 same regulatory barriers to develop their own new gen-

775 eration product). AmericanBio identified and acquired a Quadrant IV also contains projects in highly novel 825
776 company in Europe, which had the technology needed innovative areas that could break regulatory grounds and 826
777 and initial clinical data. One important criterion in their were likely to disrupt healthcare practice (Christensen, 827
CT
778 selection was a match in terms of organizational cul- 2000). The novelty of the technology, or combina- 828
779 tures. AmericanBio had earlier carried out diligence on tion of technologies, typically generated an informal 829
780 another company, but decided not to acquire it because inter-organizational web of smaller companies and col- 830
781 of ‘significant organizational differences’. Following the laborating PROs. One practitioner observed that these 831
782 acquisition, meeting timelines in product development ‘sexy technologies’ created an ‘aura of attraction’ that 832
E
783 proved difficult, one of the reasons being that the clini- drove interest and collaboration. These innovations 833
784 cal data was not as ready for FDA approval as expected, typically depended on highly networked individuals 834
RR
785 despite the fact that AmericanBio had conducted a very to orchestrate loosely coupled, decentralized projects 835
786 thorough due diligence process and, according to one and innovation relied heavily on the co-production of 836
787 respondent, ‘knew where to look for dead bodies’. knowledge across varied domains of specialist practice. 837
788 Body was based in the UK and developed its busi- Reflecting this, tasks could be described as fully, or 838
789 ness based on expertise in producing human antibodies. reciprocally, interdependent, where the sub-tasks con- 839
ANTIBODY-2 is a project concerning development of tinuously interacted because the knowledge, outputs and
CO
790 840
791 a therapeutic antibody for inflammatory disease. During decisions from one had a direct impact on the oth- 841
792 early stage clinical trials, Body decided to out-license ers, and rewards were groups-based (Thompson, 1967; 842
793 ANTIBODY-2, because they did not have the resources Wageman, 1995). 843
794 for later stage trials that would necessitate large patient DiagnosticLabs was a small US company special- 844
795 populations. The anticipation of the future partnering izing in diagnostic assays that initiated a development 845
UN
796 arrangement had a significant impact on Body’s deci- project to transform them into a ‘theragnostic’ com- 846
797 sions about the design of their early stage trials. A major pany, combining diagnostic and therapeutic products. 847
798 issue was in estimating what level of risk to accept for The logic was that availability of a targeted drug would 848
799 an expected return from the partnering arrangement that increase the market for their diagnostic and vice versa. 849
800 was yet to be agreed. The more robust the efficacy evi- This project was championed by their recently appointed 850
801 dence for ANTIBODY-2, the better the financial deal CEO, who had a reputation for managing successful 851
802 with a partner; but more trials would entail greater costs biotech companies. The choice of a specific disease as 852
+Model
853 an initial area for development, characterized by a high mercial fields, including business start-ups. Recently, 905
854 mortality rate and no approved treatment, also reflected NewTissueCo had established a commercial licensing 906
855 medical need. The new project built upon an existing deal with a larger company, BiotechCo, and had also 907
856 diagnostics kit for this disease, which was being devel- attracted further VC funding. The company was also 908
857 oped by DiagnosticLabs through collaboration with involved in a number of research collaborations with 909
858 academic partners. DiagnosticLabs lacked clinical trials TEC scientists, involving grant applications to various 910
859 and regulatory expertise and so, via the CEO’s per- UK government and international funding bodies. With 911
860 sonal networks, formed an alliance with Bioclinical, a these developments, the new CEO was able to recruit 912
F
861 company specializing in clinical trials consulting and new full-time staff and start investments in production 913
862 services, which provided a dedicated team to lead the facilities to support the licensing deal. 914
OO
863 clinical trials. The continued engagement of the lead ‘star scientists’ 915
864 The CEO used her personal connections to identify involved in setting up NewTissueCo had important rep- 916
865 a company TherapeuticCo that held IP for the matching utational and social capital effects, as well as important 917
866 therapeutic. She originally believed that this IP was sup- practical and political effects in keeping the company 918
867 ported by sufficient pre-clinical data to allow the project well connected to leading scientific teams and attuned to 919
868 to go straight into clinical trials. Bioclinical conducted the politics of the host university. Close internal connec- 920
PR
869 a due diligence assessment of TherapeuticCo’s IP on a tions within the university also lent scientific credibility 921
870 ‘good will’ basis and concluded that the preclinical data to dealings with external commercial partners—creating 922
871 available would not be sufficient to gain FDA approval a symbiosis between scientific and commercial interests. 923
872 for clinical trials and to convince VCs to provide the Importantly, networking not only involved developing 924
873 investment needed. TherapeuticCo was not interested in relationships with and through the principal scientists, 925
874 making additional investments in a non-core area and but also with researchers working within their teams, 926
ED
875 a newly appointed CEO at the company did not want to including Ph.D.s, who could potentially assist with gen- 927
876 dedicate further time to the project. In addition, Diagnos- erating new IP and commercialization. 928
877 ticLabs’s owner decided to sell the company, so halting

878 new investment. On top of this, there was a break down 5. Mechanisms linking innovation processes to 929
879 in the relationship between the CEOs of DiagnosticLabs macro capabilities 930
CT
880 and Bioclinical, as the Bioclinical team realized that

881 DiagnosticLabs was pushing for VC money for their own Tables 1 and 2 summarise the mechanisms found in 931
882 use and not for the project alliance as a whole. As a result, our cases to be important in enabling the innovation 932
883 the development project was abandoned. process for different kinds of project (recognising, as dis- 933
884 NewTissueCo was a spin-out company from TEC—a cussed, that what counted as ‘success’ was interpreted 934
E
885 leading tissue engineering research centre in the UK differently across projects), grouped in our analysis 935
886 based at a university hospital. It had preferential rights terms of how they related to integrative or relational 936
RR
887 to exploit TEC’s technologies to make ‘scaffolds’ on capabilities. 937
888 which to grow stem cells and, potentially, new organs and
889 bones. This field is extremely novel with potential appli- 5.1. Integrative capabilities 938
890 cations in the long term. NewTissueCo was established

891 by two highly regarded and experienced scientists with Firstly, our findings echo earlier work by highlight- 939
the help of the university’s business development office ing the importance of access in projects to individuals
CO
892 940
893 and seed corn funding. Although, the original intention who work ‘at the interstices’ of science and commerce 941
894 had been to find funding for the commercialization of in order to acquire relevant knowledge and expertise and 942
895 tissue-engineered products, the lack of VC interest in this to build the skills base (Powell et al., 1996; Murray, 943
896 longer-term vision led to a shift in emphasis towards the 2002; Casper and Murray, 2005). For example, in Diag- 944
897 exploitation of existing IP through licensing deals with nosticsLabs, the opportunity to develop a ‘theragnostic’ 945
UN
898 biotech companies that would enable the generation of was identified by the CEO—an experienced biomedical 946
899 revenues to fund other, riskier projects. entrepreneur who also had close networks with PROs. 947
900 Originally, the company was managed by two former Other research suggests that this mechanism is shaped 948
901 executives of large pharmas and operated as a virtual by a nation’s integrative capabilities, which reflect dif- 949
902 company employing a small number of part-time con- ferences in labour market institutions. For example, in 950
903 sultants. By 2004, a new CEO was appointed who had her study of innovation in tissue-engineered cartilage, 951
904 extensive experience in the relevant scientific and com- Murray (2002) found that the commercialization of sci- 952
+Model
Table 1
Mechanisms linked to integrative capabilities
Mechanisms Examples from cases
(1) Access to people working at interstices to acquire knowledge SampaTech: Reliance on TTO network to establish commercial contacts
and reproduction of skills base NewPharma; Reliance on ‘deal breakers’ to help commercialize the
potential product
NewTissueCo: Reproducing the scientific/commercial through training and
employing Ph.D. students
F
DiagnosticLabs: Linking to the academic community through
collaborations to develop IP and to conduct clinical trials
(2) Establishing scientific and commercial credibility in order to SampaTech: Selecting a CEO with the right profile in order to sustain
OO
ensure funding through partnering, VC or research funds connections and dialog with pharmaceutical companies
(as there are no ‘centralized’ resources available) NewPharma: Persistently trying to publish in Science in order to ‘validate’
science in the eyes of potential investors
NewTissueCo: Importance of the role of the scientific founders and host
university in providing scientific credibility
DiagnosticLabs: CEO has credibility within the VC community based on
PR
previous entrepreneurial experiences
(3) Symbolic figureheads SampaTech: No figurehead involved and so the company relied heavily on
the TTO to move the company forward
NewPharma: No figurehead scientist involved, which acted as a limiting
factor
NewTissueCo: Leading scientist’s vision, commitment and personal
experiences played a powerful role in pushing forward commercialization
ED
DiagnosticLabs: No figurehead involved implying that credibility rested
with the CEO
(4) Career perceptions and professional values in relation to SampaTech: Scientists thinking of themselves as scientists, putting
motivation to engage with innovation commercialization altruistic reasoning before commercial gain, thus constraining commercial
activity activity
NewPharma: Purely commercial interests are perceived as valid in their
CT
own right; entrepreneurial characteristics apparent

NewTissueCo: Blending of scientific, clinical and commercial orientations
in the professional identity and career choices of key individuals working
for the company
DiagnosticLabs: Medical and commercial objectives are merged together.
E
Proposed ‘theragnostic’ would allow both diagnosis and treatment of a

condition that had no previous treatment and would, at the same time,
create a bigger market by coupling sales of diagnostics with therapeutics
RR
953 ence into new medical treatments depends crucially on an important role in increasing the ‘absorptive capacity’ 967
954 overlapping, but distinctive, scientific and technologi- of our case firms, by improving their ability to recognize 968
CO
955 cal networks at the institutional level. Connections to new information, assimilate it and apply it to commer- 969
956 scientific networks, via key individuals who work at cial ends (Cohen and Levinthal, 1990). There were a 970
957 the interstices of such networks, provide ‘knowledge variety of ways in which our case companies gained 971
958 spillover’ effects that shape technological progress and access to people working at the interstices of networks: 972
959 influence firms’ abilities to develop intellectual capital through establishing links with academics (as in Diag- 973
(Zucker et al., 1998). This finding echoes earlier research nosticLabs), by enlisting ‘deal breakers’ to participate in
UN
960 974
961 on the important role of ‘boundary spanners’ who broker negotiations with potential partners (as in NewPharma), 975
962 relationships across networks (e.g. scientific and com- by training Ph.D.s who could retain links with research 976
963 mercial) and facilitate the transfer of knowledge across groups and, at the same time, assist in commercializa- 977
964 contexts (Tushman and Scanlan, 1981; Carlile, 2002). tion activities (as in NewTissueCo) and, where PROs 978
965 Having access to individuals who work at the inter- were involved, relying on a university technology trans- 979
966 stices of science, commerce and the clinic also played fer office to access networks (as in SampaTech). 980
+Model
Table 2
Mechanisms linked to relational capabilities
Mechanisms Examples from cases
(5) Alignment of interests and expectations Body: Conducting two-way due diligence with potential partners to assess
capabilities and resources but also to negotiate interests and generate mutual
understanding about expectations
AmericanBio: In deciding which company to acquire, they looked for a
‘cultural’ match, which was decisive in choosing between the two candidates.
F
Thorough due diligence, however, was not sufficient to know beforehand if
clinical data was good enough
NewTissueCo: Sought a licensing deal and investment from companies whose
OO
products, knowledge base, experiences and capabilities related closely to, but
at the same time complemented, their own
DiagnosticLabs: One way due diligence did not explore alignment of interests.
As a result, therapeutics company was not interested in further investments in
improving their IP. Further misalignment of interests occurred with the
decision of the owner to sell the company
PR
(6) Building upon existing networks to generate resources and Body: Company has a large number of diverse collaborations ranging through
sustain more risky and long term projects R and D alliances to IP licensing, that they drew upon to identify potential
partners to take the product through to development
AmericanBio: Use clinicians from their existing network to promote the
product by publishing results and increasing the body of experience and
patients
NewTissueCo: Using one business deal to leverage others becomes an
important way of enhancing commercial credibility and so building the
ED
company securing resources for long term ‘core’ project
DiagnosticLabs: Diagnostics company got involved in the project through
personal networking of its CEO. Based on the same relationship, free due
diligence was conducted on the therapeutics IP
(7) Using networks to shape regulations and ensure approval Body: Regulation expertise was considered important in selection of the
CT
potential partner, as Body would rely on it to ensure approval

AmericanBio: Company has a regulatory group dedicated to interaction with
FDA. They were able to shape the regulatory framework for the first
generation product. Next generation product approval is also being developed
through interaction with FDA
NewTissueCo: Founding scientists play an important role as opinion leaders
E
and advocates for the (emerging) discipline in scientific, business, political and
public policy arenas
DiagnosticLabs: Using its connection to bioclinical to access regulatory
RR
expertise
(8) Product ‘magnets’ Body: Therapeutic in development is for a well recognized indication meaning
relatively known path to commercialization
AmericanBio: ELBOW was the first product of its kind on the market
NewTissueCo: ‘Revolutionary’ nature of the work they are doing provides a
focus for research, but also hampers commercialization efforts
CO
DiagnosticLabs: The ‘theragnostic’ concept was new and developed over time
981 Another crucial mechanism was having members of biotechnology ventures and VC investment (McMillan 990
982 projects with scientific and commercial credibility in et al., 2000). Whilst not stressed in research on inno- 991
UN
983 order to attract VC investment and/or major research vation and networks, this dual orientation does resonate 992
984 funds. In NewTissueCo and SampaTech, for example, with research on how VC operates. Zider (1998), for 993
985 the choice of CEO with the ‘right’ scientific profile and example, found that the reputation of entrepreneurs plus 994
986 track record with investors was critical. This credibil- their business track record and ‘presentability’ to outside 995
987 ity appears to be easier to establish in the US context investors, is as, if not more, important in attracting fund- 996
988 and particularly in the Boston region (where our cases ing as their knowledge base or particular ideas; this he 997
989 focused) where there is a strong history of successful argues helps to explain US superiority in VC markets. 998
+Model
999 According to Zider (1998, p. 138): “Many entrepreneurs important (and discussed above), here we use the term 1051
1000 make the mistake of thinking that venture capitalists are ‘Figurehead’ to highlight also their symbolic and motiva- 1052
1001 looking for good ideas when, in fact, they are looking tional effects. We did not observe as many ‘figureheads’ 1053
1002 for good managers in particular industry segments”. in our US cases, nor was their importance emphasized as 1054
1003 Another way of establishing credibility for investment much by interviewees, suggesting that this mechanism 1055
1004 is via the scientific publication records of the biotech- may be particularly crucial in the UK context in bridging 1056
1005 nology firm’s scientific team (Deeds et al., 1997). In what were often more polarized values and perceptions 1057
1006 NewPharma, for example, those involved believed that of those engaged in academic, clinical and commercial 1058
F
1007 publication in a reputable scientific journal was cru- practices (Mallon et al., 2005). 1059
1008 cial for public validation. Furthermore, the affiliations This leads to the final important mechanism, which 1060
OO
1009 founders had with other organizations were also impor- concerned career perceptions and professional values of 1061
1010 tant for increasing credibility. This finding echoes work those working in PROs. There were two related issues 1062
1011 by Higgins and Gulati (2003) who demonstrated that here. The first concerned the extent to which scien- 1063
1012 the affiliations that senior managers of biotechnology tists saw their careers in both scientific and commercial 1064
1013 firms held with other organizations had an important terms—that is, as ‘entrepreneurs’ or, in Mallon et al.’s 1065
1014 symbolic value for prestigious investment banks when (2005) terms, as ‘strategic opportunists’. In Sampat- 1066
PR
1015 deciding whether to support initial public offerings. The ech, for example, scientists pursued commercialization 1067
1016 greater the prestige of the relationships, the higher the for ‘altruistic’ reasons as a means to develop science, 1068
1017 valuation – and hence private funding – a firm could whereas in New Pharma the lead scientist described him- 1069
1018 command and so the greater the likelihood of an inno- self as ‘a risk taker’ and ‘entrepreneur’. Other research 1070
1019 vation reaching development. Across our cases, such has indicated that entrepreneurial values may be less 1071
1020 affiliations appeared to be easier to establish in the US, widespread in UK academe. In a qualitative survey of 1072
ED
1021 for example, with academic scientists more frequently UK scientists in PROs, Mallon et al. (2005) found that 1073
1022 on the Scientific Boards of prestigious pharmaceuti- the majority had an over-riding sense that obvious com- 1074
1023 cal or biotech companies. This is consistent with the mercial ambition was not quite acceptable within public 1075
1024 finding that integrative capabilities encouraging move- sector science and so did not incorporate this into their 1076
1025 ment across scientific, clinical and commercial domains career planning. Only a third – described as ‘strategic 1077
CT
1026 are stronger in the US (Owen-Smith et al., 2002). The opportunists’ – were prepared to consider a move from 1078
1027 idea of a firm’s reputation-building practices provid- ‘the bench’ to a commercial career, but most of this group 1079
1028 ing legitimacy for investors is also supported in recent had become aware of the opportunities because they had 1080
1029 research by Nicholson et al. (2005), who demonstrated previous experience of working outside the public sector. 1081
1030 that biotechnology firms took a substantial discount on The second issue concerned the extent to which sci- 1082
E
1031 their first out-licensing deals with large pharmaceuti- entists perceived their professional values as scientists 1083
1032 cal firms, which was then quickly recouped through and clinicians to be aligned with, or potentially com- 1084
RR
1033 significantly higher valuations from VC at subsequent promised by, the pursuit of commercial activity. In our 1085
1034 financing rounds. cases, the actual movement of scientists from academe 1086
1035 The importance of reputation building and the sym- to commerce was rare. In all the cases, where PROs 1087
1036 bolic value of affiliations relates closely to another were involved, the lead scientists remained as scientists, 1088
1037 mechanism identified in our cases, which was the and recruited other people to push forward commer- 1089
symbolic role played by ‘Figureheads’—usually lead cialization. However, there were important differences,
CO
1038 1090
1039 scientists with international reputations and commer- reflected in the professional values of scientists across 1091
1040 cial and/or clinical experience. Such individuals, when UK and US contexts. So, in NewPharma and Diagnos- 1092
1041 linked closely to projects, played a key role in mobilizing ticLabs, for example, commercialization was seen as 1093
1042 support amongst diverse users and clinicians. For exam- meaningful in its own right, both as a means to clini- 1094
1043 ple, the lead scientist in NewTissue embodied (quite cal improvement and as a way of generating personal 1095
UN
1044 literally, having been a former patient himself as well as a revenue for the scientists. In contrast, in SampaTech 1096
1045 lead clinician) the overall vision of the project, symboliz- and NewTissueCo, the major motivation was to ‘make 1097
1046 ing its clinical and scientific significance to user groups. a real difference to patients’ lives’ and to use this activ- 1098
1047 The importance of ‘star scientists’ has been noted else- ity to help fund continued scientific development, not to 1099
1048 where, but usually in terms of the role such individuals make their personal fortune. This suggests that, at least 1100
1049 play in linking commercial activity to the knowledge in our cases, the mobility of individuals’ careers across 1101
1050 base of academe (Zucker et al., 1998). Whilst this was boundaries of science and commerce emphasized else- 1102
+Model
1103 where (Casper and Murray, 2005) was less significant this was the due diligence process. Where due diligence 1153
1104 in incentivizing (or de-incentivizing) commercialization was two-way (conducted by each party involved), as in 1154
1105 than the professional values that individuals attached the Body case, projects were able to balance partner 1155
1106 to their work. One of the problems in the UK is that interests and commitments. Moreover, once a collabo- 1156
1107 individuals’ career perceptions are still quite polarized ration agreement was in place, it was important to have 1157
1108 and scientific entrepreneurship less strongly advocated mechanisms to ensure continuous monitoring and allow 1158
1109 (Clark, 2000; Turpin and Deville, 1995; Mallon et al., early identification of potential misalignment of inter- 1159
1110 2005). ests or expectations. This was achieved, for example in 1160
F
1111 Finally, our analysis suggested that these mechanisms the Body and AmericanBio cases, by matching project 1161
1112 relating to integrative capabilities were more crucial in structures and processes to partners so that issues and 1162
OO
1113 projects characterised by loose, rather than tight, orga- concerns could quickly be identified. Where this was 1163
1114 nizational coupling. This can be explained in terms of not achieved, as in the DiagnosticLabs case, the col- 1164
1115 two major impacts of a nation’s integrative capabilities. laboration quickly broke down. Other research has also 1165
1116 First, such capabilities have an impact on the motiva- identified that the alignment of interests and expecta- 1166
1117 tion of scientists to engage in commercialization, moving tions between partners is crucial for innovation involving 1167
1118 between academia and industry and connecting basic R collaboration between biotechnology firms and pharma- 1168
PR
1119 and D. Thus, mechanisms 1 and 4 were important ways ceutical firms (Rhodes et al., 2003). As Rhodes et al. put 1169
1120 in which this motivation was facilitated. They helped it: “the process of partner identification should not be 1170
1121 to orchestrate loose networks characterising projects in undertaken opportunistically” (p. 300). 1171
1122 quadrants I and IV and made possible different kinds of A second mechanism linked to relational capabilities 1172
1123 transactions (e.g. between key scientists, entrepreneurs that was important at the project level was the ability 1173
1124 and investors). Second, such capabilities have an impact to build upon existing inter-organizational networks to 1174
ED
1125 on the ability to secure resources (e.g. from research generate resources and buy-in from users (i.e. clinicians). 1175
1126 funding, VC and through partnerships) needed to move This links to recent literature on innovation ecosystems 1176
1127 innovation forward in situations where these resources (Adner, 2006) that identifies the importance of build- 1177
1128 are not available from and cannot be centrally controlled ing markets as well as ‘brilliant products’. Moreover, 1178
1129 by one organization. Mechanisms 2 and 3 were impor- it is clear that users can contribute to the development 1179
CT
1130 tant in this respect. In tightly coupled innovation projects, of the product itself (Lettl et al., 2006; Von Hippel, 1180
1131 there was less need to rely on such mechanisms because 2005). Enrolment (of users and resources) in our cases 1181
1132 the R and D was controlled from within larger biotech- involved building from existing networks which then 1182
1133 nology or pharmaceutical firms, which deployed internal acted as a ‘centre of gravity’ for further networking and 1183
1134 resources and were populated by scientists who, given enrolment (Kreiner and Schultz, 1993). For example, 1184
E
1135 that they worked in a commercial environment, were AmericanBio’s acquisition of a European company gave 1185
1136 directly incentivized to commercialize basic research. it access, not only to that company’ technology, but also 1186
RR
1137 In this way, the analysis indicates that a nation’s inte- to the network of influential clinicians and clinical data 1187
1138 grative capabilities might be expected to have relatively it had established. AmericanBio also committed signif- 1188
1139 stronger effects on projects in quadrants I and IV. Inter- icant resources to training clinicians in the new surgical 1189
1140 estingly, these mechanisms were not simply focused on techniques so that they could act as ‘opinion leaders’ for 1190
1141 knowledge integration but, rather, were most important their product in the clinical community (Rogers, 1995). 1191
in terms of building the relationships across the aca- NewTissueCo also capitalized on existing relationships
CO
1142 1192
1143 demic/commercial communities in order to establish the by closing licensing deals and using these to leverage 1193
1144 reputation and claims to knowledge required to attract commercial viability and secure resources for their more 1194
1145 investment. long term and riskier core project. Furthermore, Diag- 1195
nosticLabs used their CEO’s existing personal networks 1196
1146 5.2. Relational capabilities to conduct due diligence for free from Bioclinical, with 1197
UN
the promise of future gains should the collaboration 1198
1147 Turning to relational capabilities, these concern how prove fruitful. 1199
1148 easy/difficult it is in the macro institutional context A third important mechanism was interaction with 1200
1149 to establish and maintain collaborations across diverse regulators, either directly to shape regulation and ensure 1201
1150 organizations. At project level, alignment of interests and approval (as in AmericanBio), or with other firms 1202
1151 expectations was a crucial mechanism in our cases and that had regulatory expertise and could help develop 1203
1152 can be linked to relational capabilities. Key to achieving required documentation and interactions with the FDA 1204
+Model
1205 (as in Body and DiagnosticsLabs). Given the uncer- offices or investors) their main challenge was to develop 1257
1206 tainty associated with treatments being developed in their credibility across the scientific/commercial divide; 1258
1207 biotech companies, such collaborations were extremely knowledge integration per se was less important. Inter- 1259
1208 important, with regulators (and regulations) acting as estingly, then, the mechanisms we identified that we 1260
1209 ‘obligatory passage points’ in networks (Callon, 1986). argue link to relational capabilities were more focused 1261
1210 Many small biotech firms do not have this expertise on ways in which relationships could foster knowledge 1262
1211 (Hawthorne, 2005) and so are very dependent on spe- integration by overcoming pragmatic boundaries, rather 1263
1212 cialized consultants to provide it. than on building the relationships per se (Carlile, 2004). 1264
F
1213 Finally, our cases demonstrated the importance of
1214 having in projects a product ‘magnet’ to bridge orga- 6. General discussion and conclusions 1265
OO
1215 nizations in the market place (Doz et al., 2001). Thus,
1216 in NewTissueCo, the ‘revolutionary’ nature of the inno- This study has explored how macro-level capabilities 1266
1217 vation mobilized different stakeholders in the network relate to micro biomedical innovation processes, focus- 1267
1218 and helped entice other organizations to want to work ing on integrative and relational capabilities found to 1268
1219 with the company on this ‘leading edge’ technology. In be important in the biomedical domain (Owen-Smith 1269
1220 AmericanBio, ELBOW was a well developed but inno- et al., 2002). This builds from earlier research that has 1270
PR
1221 vative product, where desired characteristics for the next shown how institutions governing labor, finance and 1271
1222 generation were known in the user and business commu- product markets affect innovation activities and the per- 1272
1223 nities. Similarly, in DiagnosticLabs, the ‘theragnostic’ formance of sectors and nations (Nelson, 1993; Hall and 1273
1224 concept helped bring together different ‘pieces of the Soskice, 2001; Clark, 1987). A main contribution has 1274
1225 puzzle’. When there was no clear product ‘magnet’ – been to identify and unpack mechanisms (depicted in 1275
1226 for example, in cases with multiple possible indications Tables 1 and 2) that appear crucial in helping to explain 1276
ED
1227 derived from the same compound – the project could how macro-level capabilities play out at the level of 1277
1228 suffer from a lack of focus. Whilst product magnets are micro-level innovation projects. These mechanisms are 1278
1229 perhaps more closely linked to markets than networks, not exhaustive but provide a useful starting point in terms 1279
1230 our data imply that they may help to generate activity of understanding the processes through which macro 1280
1231 and networking around particular areas of innovation, capabilities may come to influence innovation processes. 1281
CT
1232 so potentially shaping (and being shaped by) relational It is worth noting that the mechanisms identified 1282
1233 capabilities. played a role in shaping innovation processes but may 1283
1234 Our analysis suggested that these mechanisms were also be shaped by them. For example, having scientific 1284
1235 more crucial in those projects characterized by high and commercial credibility with investors both influ- 1285
1236 knowledge boundaries (quadrants III and IV). This enced innovation processes (by attracting funding) but 1286
E
1237 appeared to be because of the highly novel nature of was also influenced by them (past ‘successes’ gener- 1287
1238 the innovations in these projects and the very long inno- ating credibility, for example). Moreover, whilst we 1288
RR
1239 vation life cycles. This meant that there was a need did not address this in our study, these mechanisms 1289
1240 to constantly leverage resources in order to support might influence, as well as be influenced by institu- 1290
1241 long term commercial developments. Moreover, for a tional arrangements and capabilities. For example, prior 1291
1242 project to be ‘successful’, it needed to deal with prob- research has found that labor markets and integrative 1292
1243 lems associated with the disruption to existing practices capabilities at the macro level influence individuals’ 1293
that these innovations could introduce. Our case projects career perceptions and mobility, but that this relation-
CO
1244 1294
1245 used and built upon existing networks to generate and ship is recursive since they are also influenced by them 1295
1246 sustain resources and ‘buy-in’ (mechanism 6) and to (Casper and Murray, 2005). Moreover, these mecha- 1296
1247 ensure approval (mechanism 7). Furthermore, the nature nisms work together in innovation processes, not in 1297
1248 of these innovations meant that knowledge integration isolation. For example, in NewTissueCo a strong figure- 1298
1249 in the context of complex inter-organizational dynam- head role was coupled with complementary commercial 1299
UN
1250 ics was crucial. Such projects encountered, in Carlile’s expertise and a relatively strong product magnet, which 1300
1251 (2004) terms, political or ‘pragmatic’ boundaries. Align- allowed the project to secure further funding, despite the 1301
1252 ing interests and expectations (mechanism 5) and having overriding suspicion of commercially oriented activity 1302
1253 a strong product magnet (mechanism 8) was helpful by the UK founder. 1303
1254 in terms of dealing with such pragmatic boundaries. Where previous work has focused on benefits of 1304
1255 While projects in quadrants I also involved collabora- macro capabilities in terms of knowledge flows, our 1305
1256 tion with other organizations (e.g. technology transfer data suggest that issues of credibility, values and per- 1306
+Model
1307 ceptions, and alignments of interests are equally, if not

1308 more, important in terms of mobilizing innovation at
1309 the project level. Yet, most policy initiatives aimed at
1310 encouraging innovation in life sciences (e.g. technology
1311 transfer, network grants, joint patents and so forth) focus
1312 on knowledge flows/knowledge transfer between public
1313 and private organizations. Our study suggests that the
1314 effects of these initiatives may be limited if attention
F
1315 is not also given to the normative mechanisms high-
1316 lighted here. In the UK context, where scientific, clinical
OO
1317 and commercial interests are more clearly demarcated
1318 (Clark, 2000), these latter, normative, concerns become
1319 even more central than in the US, where the hybridization
1320 of scientific, clinical and commercial values is generally
1321 more acceptable (Clark, 2000). To extend a construction
1322 metaphor, in the UK, building bridges allow knowledge Fig. 2. The importance of integrative and relational capabilities for
PR
1323 to flow between PROs, commercial and clinical orga- modes of organizing biomedical innovation.
1324 nizations may be problematic because the ends of the
1325 bridge may be substantively different. part of Fig. 2 (quadrant II) are less affected by macro 1359
1326 A second contribution is to consider the relative capabilities and so we may not expect significant dif- 1360
1327 importance of macro capabilities at the institutional level ferences across US and UK contexts for this mode. In 1361
1328 for characteristically kinds of innovation projects. Thus, quadrant II, innovation projects are dominated by large 1362
ED
1329 we can propose that the inﬂuence of institutionalized pharmaceutical firms, which operate on a global basis. 1363
1330 capabilities on innovation process at the micro level is Arguably, then, national institutional contexts might play 1364
1331 systematically related to different modes of organizing a less important role here, as compared with the features 1365
1332 innovation. Whilst our inductive methodology allows us of the particular lead organizations, which have a rela- 1366
1333 to generate this proposition, future research would be tively high degree of autonomy and control (Hardy and 1367
CT
1334 needed to confirm (or refute) it. However, our analysis Phillips, 1998; Hardy et al., 2003). 1368
1335 suggests, in line with other theorists, that institutional On the other hand, projects characterized by the upper 1369
1336 arrangements do not determine innovation processes but right part of Fig. 2 (quadrant IV) are likely to be most 1370
1337 may, as Clark (2000) puts it, generate ‘zones of manoeu- affected by macro capabilities of the particular institu- 1371
1338 vre’ that allow some kinds of activities to occur more tional context in which they operate. Here, mechanisms 1372
E
1339 easily than others. Moreover, it suggests that the relative relating to both integrative and relational capabilities 1373
1340 disadvantages of being in an institutional context (the appear to be important in order to orchestrate loose 1374
RR
1341 UK) that is less supportive of integrative and relational relationships and bridge knowledge boundaries, mean- 1375
1342 capabilities can be overcome where mechanisms of the ing that stronger differences in innovation performance 1376
1343 kinds we have identified can nevertheless be developed would be expected between the UK and US for these 1377
1344 (in NewTissue, for example). Similarly, whilst Mallon kinds of projects. While we do not have comparative 1378
1345 et al.’s (2005) study of UK scientists in PROs found data per se that would allow us to confirm that these 1379
the majority to be uncomfortable with coupling scien- types of project were less frequent and less successful
CO
1346 1380
1347 tific and commercial ambition, a significant minority (the in the UK as compared to the US, our 1st phase sur- 1381
1348 ‘strategic opportunists’) were prepared to move between vey data did indicate that a greater instance of quadrant 1382
1349 ‘the bench’ and a commercial career. IV type innovation projects in the US than in the UK, 1383
1350 This suggests that generic statements about relative and more problems associated with such projects in the 1384
1351 national advantage in biomedical innovation need to be UK context. Whilst, the case of NewTissue showed that 1385
UN
1352 tempered by a consideration of the kinds of project and quadrant IV type projects in the UK could survive, we 1386
1353 the combination of mechanisms deployed at project, firm would predict that it would take considerably more effort 1387
1354 or sector levels (Casper and van Waarden, 2005). The to develop appropriate mechanisms at project level to 1388
1355 framework depicted in Fig. 2 attempts, then, to provide cope with the challenges of networking across the aca- 1389
1356 further insights into the relative advantage/disadvantage demic/commercial divide and building organizational 1390
1357 of particular national contexts for innovation. Thus, our collaborations in this context. We would anticipate, then, 1391
1358 analysis suggests that innovation processes in lower left that quadrant IV type projects would be more problem- 1392
+Model
1393 atic in the UK context and might move more rapidly demarcated ‘discipline’, then innovation projects could 1445
1394 than their US counterparts, to another quadrant where the shift to quadrants with characterised as low in knowledge 1446
1395 macro-institutional context has less of a divisive impact. boundaries. Third, whilst our classification of mecha- 1447
1396 Again, this is something that future empirical research nisms in terms of whether they link to integrative or 1448
1397 could more directly evaluate. relational capabilities is useful for analytical purposes, 1449
1398 In the middle part of Fig. 2 (quadrants I and III), we clearly integrative and relational capabilities are them- 1450
1399 expect moderate effects of the macro capabilities—they selves related concepts (both being linked to network 1451
1400 matter but perhaps not enough to produce systemati- ties, for example). Thus, feasibly, some mechanisms 1452
F
1401 cally large differences across the US and UK contexts. might relate to both capabilities to a greater or lesser 1453
1402 For example, for quadrant I, where integrative capability extent. Further work would be required to tease our inter- 1454
OO
1403 has more importance, specific initiatives can be built to relationships and test further our initial propositions. 1455
1404 cope with the more rigid divide between academia and Turning finally to policy implications, this study sug- 1456
1405 industry in the UK context, for example by introduc- gests a need for national policy aimed at improving 1457
1406 ing policies that direct TTO’s activities to assist spin-off biomedical innovation to be sensitive to the different 1458
1407 companies and scientists in their participation within ways of organizing innovation identified here. Taking 1459
1408 loose networks (as described in the case examples). In these implications further, it suggests potential perverse 1460
PR
1409 quadrant III, where relational capability has more impor- effects of, supposedly supportive, policy initiatives for 1461
1410 tance, innovation projects are conducted within larger knowledge transfer. For example, in the UK, government 1462
1411 organizations, and of course some UK-based firms can policies aimed at helping academic scientists to switch to 1463
1412 be as experienced as US-based firms in developing the industry careers (e.g. by starting up businesses) or uni- 1464
1413 required collaborations. versity policies which allow academics to engage in a 1465
1414 This kind of analysis extends existing research that specified number of days’ consultancy ‘outside’ of their 1466
ED
1415 treats the impact of integrative and relational capabili- academic work, may actually serve to reinforce the fun- 1467
1416 ties on innovative performance in the biomedical sector damental gap between academic and commercial values 1468
1417 as, in broad terms, uniformly positive (Owen-Smith and career interests. More important in the UK context, 1469
1418 et al., 2002), by suggesting that their impact may be might be to develop more normatively-based initiatives 1470
1419 more acute for some kinds of project than others. Our and incentives that encourage, for want of a better term, 1471
CT
1420 study also helps to address a central critique of com- ‘strategic opportunism’ to become a more legitimate part 1472
1421 parative institutional studies concerning, as Casper and of academic practice (Mallon et al., 2005). Such initia- 1473
1422 Murray (2005; 56) put it (in relation to labor mar- tives might include, for example, industry secondments 1474
1423 ket institutions), “the limited connection made between of doctoral students or incentives for academic scientists 1475
1424 macro-institutions and the micro-dynamics (of individ- to participate in commercial scientific advisory boards. 1476
E
1425 ual careers) through which these institutional differences Finally, our results suggest that attempts to replicate 1477
1426 are manifested”. US policy are unlikely to be fully effective in the UK 1478
RR
1427 Before discussing the implications for policy, there context. For example, the UK Department of Trade and 1479
1428 limitations that should be dealt with. First, this is an Industry (DTI), together with UK research councils, are 1480
1429 exploratory study using a limited number of firms and keen to replicate ‘MIT-type’ institutional mechanisms 1481
1430 projects; therefore, we cannot generalize our findings to support innovation research (e.g. through initiatives 1482
1431 to the larger population of firms and projects. Whilst such as the Cambridge-MIT Institute and the ‘Innova- 1483
we can suggest contingencies between the impact of tion Challenge’). However, as seen, the UK context may
CO
1432 1484
1433 macro capabilities and different kinds of projects, the not support the hybridization of professional and occu- 1485
1434 propositions developed here require further research on pational practices seen in the US, or the development of 1486
1435 a broader sample of development projects in the UK and scientific entrepreneurs, required to make such a model 1487
1436 US. Second, whilst our framework is useful in identi- work. 1488
1437 fying patterns of innovation, it is clearly broadly-based

UN
1438 and more fine-tuned analysis of particular modes would

1439 no doubt be able to identify further variation in the orga- References 1489
1440 nization of biomedical innovation (Lockett et al., 2005).

Adner, R., 2006. Match your innovation strategy to your innovation 1490
1441 It should also be noted that innovation processes are not
ecosystem. Harvard Business Review 84 (4), 98–107. 1491
1442 confined to particular ‘boxes’—they may shift modes Aldrich, H., 2000. Organizations Evolving. Sage, London. 1492
1443 across their lifecycle. For example, if tissue engineer- Alter, C., Hage, J., 1993. Organizations Working Together. Sage, New- 1493
1444 ing does eventually establish itself as a professionally bury Park, CA. 1494
+Model
1495 Alvesson, M., Skoldberg, K., 2000. Reflexive Methodology: New Vis- Gay, B., Dousset, B., 2005. Innovation and network structural dynam- 1557
1496 tas for Qualitative Research. Sage Publications, London. ics: study of the alliance network of a major sector of the 1558
1497 Bock, A.-K., Ibarreta, D., Rodriguez-Cerezo, E., 2003. Human biotechnology industry. Research Policy 34, 1457–1475. 1559
1498 tissue-engineered products: today’s markets and future prospects. Gibbons, M., Limoges, C., Nowotny, H., Schwartzman, S., Scott, P., 1560
1499 Report published by the Institute for Prospective Technological Trow, M., 1994. The New Production of Knowledge: The Dynam- 1561
1500 Studies. ics of Science and Research in Contemporary Societies. Sage, 1562
1501 Callon, M., 1986. Some elements in a sociology of translation: domes- London. 1563
1502 tication of the scallops and fisherman of St. Brieuc Bay. In: Law, Gittell, J.H., Weiss, L., 2004. Coordination networks within and across 1564
1503 J. (Ed.), Power, Action and Belief. Routledge, London. organizations: a multi-level framework. Journal of Management 1565
F
1504 Carlile, P.R., 2004. Transferring, translating, and transforming: an inte- Studies 41 (1), 127–153. 1566
1505 grative framework for managing knowledge across boundaries. Glaser, B., 1978. Theoretical Sensitivity: Advances in the Methodol- 1567
1506 Organization Science 5 (5), 555–568. ogy of Grounded Theory. Sociology Press. 1568
OO
1507 Carlile, P., 2002. A pragmatic view of knowledge and boundaries: Hall, P.A., Soskice, D. (Eds.), 2001. Varieties of Capitalism: The 1569
1508 boundary objects in new product development. Organization Sci- Institutional Foundations of Comparative Advantage. Oxford Uni- 1570
1509 ence 13, 442–455. versity Press, Oxford. 1571
1510 Carlsson, B. (Ed.), 2002. Technological Systems in the Bio Industries: Hardy, C., Phillips, N., 1998. Strategies of engagement: lessons from 1572
1511 An International Study. Kluwer Academic Publishers, Boston. the critical examination of collaboration and conflict in an interor- 1573
1512 Casper, S., 2000. Institutional adaptiveness, technology policy, and the ganizational domain. Organization Science 9 (2), 217–230. 1574
1513 diffusion of new business models: the case of German biotechnol- Hardy, C., Phillips, N., Lawrence, T., 2003. Resources, knowledge and 1575
PR
1514 ogy. Organization Studies 21, 887–914. influence: the organizational effects of interorganizational collab- 1576
1515 Casper, S., Kettler, H., 2001. National institutional frameworks and oration. Journal of Management Studies 40 (2), 321–346. 1577
1516 the hybridization of entrepreneurial business models: The Ger- Hawthorne, F., 2005. Inside the FDA: The Business and Politics Behind 1578
1517 man and UK biotechnology sectors. Industry and Innovation 8, the Drugs that We Take and the Food that We Eat. Wiley, Hoboken, 1579
1518 5–30. NJ. 1580
1519 Casper, S., Murray, F., 2005. Exchange under conflicting institutional Henderson, R., 1994. The evolution of integrative capability: inno- 1581
1520 logics: commerce and science in the world of population genetics. vation in cardiovascular drug discovery. Industrial and Corporate 1582
ED
1521 Unpublished manuscript. Change 3 (3), 607–630. 1583
1522 Casper, S., van Waarden, F. (Eds.), 2005. Innovation and Institutions: Higgins, M., Gulati, R., 2003. The effects of upper echelon affiliations 1584
1523 A Multidisciplinary Review of the Study of Innovation Systems. on underwriter prestige. Organization Science 14 (3), 244–263. 1585
1524 New Horizons of the Economics of Innovation series. Edward Elgar Hilton, T., Flanzer, J., Cartwright, W., Fletcher, B., 2002. Resistance 1586
1525 Publishing, Cheltenham. to innovation among US drug abuse treatment providers: When 1587
1526 Christensen, C.M., 2000. The Innovator’s Dilemma. Harper Business, organizational knowledge interferes with organizational learning. 1588
CT
1527 New York. In: Paper presented at the Organizational Knowledge, Learning and 1589
1528 Clark, P., 1987. Anglo-American Innovation. De Gruyter, New York. Capabilities Conference, Athens. 1590
1529 Clark, P., 2000. Organizations in Action: Competition Between Con- Howard-Grenville, J.A., Carlile, P. The incompatibility of knowledge 1591
1530 texts. Routledge, London. regimes: consequences of the material world for cross-domain 1592
1531 Clark, P., 2003. Organizational Innovations. Sage, London. work. European Journal of Information Systems, in press. 1593
1532 CMR International, 2006. Centre for Medicines Research International Kreiner, K., Schultz, M., 1993. Informal collaboration in R and D: the 1594
E
1533 Pharmaceutical R and D Factbook. CMR International, Surrey. formation of networks across organizations. Organization Studies 1595
1534 Cohen, W.M., Levinthal, D.A., 1990. Absorptive capacity: a new 14 (2), 189–209. 1596
1535 perspective on learning and innovation. Administrative Science Lettl, C., Herstatt, C., Gemuenden, H.G., 2006. Users’ contribution to 1597
RR
1536 Quarterly 35 (1), 128–152. radical innovation: evidence from four cases in the field of medical 1598
1537 Coombs, R., Harvey, M., Tether, B.S., 2003. Analyzing distributed equipment technology. R and D Management 36 (3), 251–272. 1599
1538 processes of provision and innovation. Industrial and Corporate Liebeskind, J.P., Oliver, A.L., Zucker, L., Brewer, M., 1996. Social 1600
1539 Change 12 (6), 1125–1155. networks, learning and flexibility: souring scientific knowledge in 1601
1540 Deeds, D.L., Decarolis, D., Coombs, J., 1997. The impact of firm- new biotechnology firms. Organization Science 7 (4), 428–443. 1602
1541 specific capabilities on the amount of capital raised in an initial Lockett, A., Siegel, D., Wright, M., Ensley, M.D., 2005. The creation 1603
CO
1542 public offering: evidence from the biotechnology industry. Journal of spin-off firms at public research institutions: managerial and 1604
1543 of Business Venturing 12 (1), 31–46. policy implications. Research Policy 34, 981–993. 1605
1544 Denzin, N., Lincoln, Y., 1998. Entering the field of qualitative research. Mallon, M., Duberley, J., Cohen, L., 2005. Careers in public sector 1606
1545 In: Denzin, N., Lincoln, Y. (Eds.), Strategies of Qualitative Inquiry. science: orientations and implications. R and D Management 35 1607
1546 Sage, London, pp. 1–34. (4), 395–407. 1608
1547 Dodgson, M., Gann, D., Salter, A., 2004. Innovation technology: McMillan, G.S., Narin, F., Deeds, D., 2000. An analysis of the critical 1609
1548 exploring the impact of simulations on the innovation process. role of public science in innovation: the case of biotechnology. 1610
UN
1549 Paper presented at DRUID summer conference. Industrial Dynam- Research Policy 29, 1–8. 1611
1550 ics, Innovation and Development, Helsingore, Denmark. Miles, M.B., Huberman, A.M., 1994. Qualitative Data Analysis. Cam- 1612
1551 Dopson, S., 2005. The diffusion of medical innovations: can con- bridge University Press, Cambridge. 1613
1552 figurational sociology contribute? Organization Studies 26 (8), Murray, F., 2002. Innovation as co-evolution of scientific and tech- 1614
1553 1125–1144. nological networks: exploring tissue engineering. Research Policy 1615
1554 Doz, Y., Santos, J., Williamson, P., 2001. From Global to Metana- 31, 1389–1403. 1616
1555 tional: How Companies Win in the Knowledge Economy. Harvard Nelson, R. (Ed.), 1993. National Innovation Systems: A Comparative 1617
1556 Business School Press, Boston. Analysis. Oxford University Press, New York/Oxford. 1618
+Model
1619 Nicholson, S., Danzon, P., McCullough, J., 2005. Biotech-pharma- Saunders, M., Lewis, P., Thornhill, A., 2000. Research Methods for 1655
1620 ceutical alliances as a signal of asset and firm quality. Journal of Business Students. Prentice Hall, FT. 1656
1621 Business 78 (4), 1433–1464. Stake, R.E., 1995. Case studies. In: Denzin, N.K., Lincoln, Y.S. (Eds.), 1657
1622 Nowotny, H., Scott, P., Gibbons, M., 2001. Re-Thinking Science: Handbook of Qualitative Research. Sage, Thousand Oaks, CA, pp. 1658
1623 Knowledge and the Public in an Age of Uncertainty. Polity, Cam- 236–247. 1659
1624 bridge. Stokes, D.E., 1997. Pasteurs Quadrant: Basic Science and Technolog- 1660
1625 Orlikowski, W.J., 2002. Knowing in practice: enacting a collective ical Innovation. Brookings. 1661
1626 capability in distributed organizing. Organization Science 13 (3), Swan, J., Bresnen, M., Mendes, M., Newell, S., Perkmann, M., Rober- 1662
1627 249–273. ston, M., 2005. Exploring interactivity in biomedical innovation: 1663
F
1628 Owen-Smith, J., 2003. From separate systems to a hybrid order: accu- a framework and case study analysis. In: Proceedings of the 5th 1664
1629 mulative advantage across public and private science at Research European Conference on Organizational Knowledge, Learning and 1665
1630 One universities. Research Policy 32, 1081–1104. Capabilities, Boston, March. 1666
OO
1631 Owen-Smith, J., Riccaboni, M., Pammolli, F., Powell, W., 2002. A Thompson, J., 1967. Organizations in Action. McGraw-Hill, New 1667
1632 comparison of US and European university–industry relations in York. 1668

1633 the life sciences. Management Science 48 (1), 24–43. Turpin, T., Deville, A., 1995. Occupational roles and expectations of 1669
1634 Owen-Smith, J., Powell, W.W., 2004. Knowledge networks as channels research scientists and research managers in scientific research 1670
1635 and conduits: the effects of spillovers in the Boston Biotechnology institutions. R and D Management 25 (2), 141–157. 1671
1636 Community. Organization Science 15 (1), 5–22. Tushman, M., Scanlan, T., 1981. Boundary spanning individuals: their 1672
1637 Pettigrew, A.M., 1990. Longitudinal field research on change: theory role in information transfer and their antecedents. Academy of 1673
PR
1638 and practice. Organization Science 1 (3), 267–292. Management Journal 24, 289–305. 1674
1639 Powell, W., Koput, W., Smith-Doerr, L., 1996. Interorganizational Tsoukas, H., 1989. The validity of idiographic research explanations. 1675
1640 collaboration and the locus of innovation: networks of learn- Academy of Management Review 14 (4), 551–561. 1676
1641 ing in biotechnology. Administrative Science Quarterly 41 (1), Von Hippel, E., 2005. Democritizing Innovation. The MIT Press, 1677
1642 116–130. Boston. 1678
1643 Rasmussen, B., February 2005. Pharmaceutical industry project. Wageman, R., 1995. ‘Interdependence and group effectiveness’. 1679
1644 Working Paper Series. Centre for Strategic Economic Studies, Administrative Science Quarterly, 145–180. 1680
ED
1645 Victoria. University of Technology. Whitley, R., 2000. The institutional structuring of innovation strategies: 1681
1646 Rhodes, I., Nelson, C., Berman, G., 2003. The key to successful col- business systems, firm types and patterns of technical change in 1682
1647 laborations: rigorous and independent due diligence. Journal of different market economies. Organization Studies 21 (5), 855–886. 1683
1648 Commercial Biotechnology 9 (4), 297–305. Whitley, R., 2003. Competition and pluralism in the public sciences: 1684
1649 Rogers, E.M., 1995. Diffusion of Innovations, fourth ed. Free Press, the impact of institutional frameworks on the organization of aca- 1685
1650 New York. demic science. Research Policy 32 (6), 1015–1029. 1686
CT
1651 Rosenberg, N., Nelson, R., 1994. American universities and technical Zider, B., 1998. How venture capital works. Harvard Business Review 1687
1652 advance in industry. Research Policy 23, 323–348. 76 (6), 131–139. 1688
1653 Salman, N., Saives, A.-L., 2005. Indirect networks: an intangible Zucker, L., Darby, M., Brewer, M., 1998. Intellectual capital and 1689
1654 resource for biotechnology innovation. R and D Management 35 the birth of US biotechnology enterprises. American Economic 1690
(2), 203–215. Review 88, 290–306. 1691

E
RR
CO
UN

Research Policy 2007

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Research Policy 2007

Uploaded by

Copyright:

Available Formats

+Model

RESPOL 2007 1–19 ARTICLE IN PRESS

Research Policy xxx (2007) xxx–xxx

3 Modes of organizing biomedical innovation in the UK and

20 © 2007 Published by Elsevier B.V.

21 Keywords: Integrative capabilities; Relational capabilities; Innovation; Organization; Biomedical

23 1. Introduction ogy to radically change (and hopefully improve) medical 34

treatments and diagnostic techniques, is high. However, 35

27 knowledge to improve the delivery of human healthcare International, 2006). 39

often cut across established professional, occupational 46

and organizational boundaries, and threaten to dis- 47

∗ Corresponding author. Tel.: +44 2476524271;

1 0048-7333/$ – see front matter © 2007 Published by Elsevier B.V.

69 shown how systematic variation in the composition of at project level. 121

167 be identified as relevant across contexts, then we could 2002). 217

173 future research. in the US is qualitatively different to Europe, embrac- 223

ing more diverse constituents and knowledge sources, 224

284 enhanced likelihood of ‘spillover effects’: that is, knowl-

286 (Owen-Smith and Powell, 2004; Murray, 2002; Kreiner

299 regions. who had significant experience of working in early-stage 349

469 described. This included, for example, access to finance,

497 anisms as related to macro capabilities.

503 these into 8 primary mechanisms identified from the 10

508 researchers on the project (Denzin and Lincoln, 1998).

582 ment (Howard-Grenville and Carlile, in press; Gay and

584 simply multidisciplinarity, but whether or not the work

588 come. companies started by academics with entrepreneurial 638

674 They delivered presentations to senior pharmaceutical

676 ceuticals. Where individuals appeared to be interested,

772 shaping US regulation, which made it difficult for com-

774 same regulatory barriers to develop their own new gen-

877 ticLabs’s owner decided to sell the company, so halting

880 and Bioclinical, as the Bioclinical team realized that

887 to exploit TEC’s technologies to make ‘scaffolds’ on capabilities. 937

890 cations in the long term. NewTissueCo was established

own right; entrepreneurial characteristics apparent

Proposed ‘theragnostic’ would allow both diagnosis and treatment of a

potential partner, as Body would rely on it to ensure approval

nosticLabs used their CEO’s existing personal networks 1196

the promise of future gains should the collaboration 1198

1307 ceptions, and alignments of interests are equally, if not

1437 fying patterns of innovation, it is clearly broadly-based

1438 and more fine-tuned analysis of particular modes would

1440 nization of biomedical innovation (Lockett et al., 2005).

1518 5–30. NJ. 1580

1632 comparison of US and European university–industry relations in York. 1668

1642 116–130. Boston. 1678

(2), 203–215. Review 88, 290–306. 1691

You might also like